Conditional random pattern algorithm for LOH inference and segmentation |
| |
| Authors: | Wu Ling-Yun Zhou Xiaobo Li Fuhai Yang Xiaorong Chang Chung-Che Wong Stephen T C |
| |
| Institution: | 1Center for Biotechnology and Informatics, Department of Radiology, The Methodist Hospital Research Institute, Weill Medical College, Cornell University, Houston, TX 77030, USA, 2Institute of Applied Mathematics, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100080, China and 3Department of Pathology, The Methodist Hospital, Weill Medical College, Cornell University, Houston, TX 77030, USA |
| |
| Abstract: | Motivation: Loss of heterozygosity (LOH) is one of the mostimportant mechanisms in the tumor evolution. LOH can be detectedfrom the genotypes of the tumor samples with or without pairednormal samples. In paired sample cases, LOH detection for informativesingle nucleotide polymorphisms (SNPs) is straightforward ifthere is no genotyping error. But genotyping errors are alwaysunavoidable, and there are about 70% non-informative SNPs whoseLOH status can only be inferred from the neighboring informativeSNPs. Results: This article presents a novel LOH inference and segmentationalgorithm based on the conditional random pattern (CRP) model.The new model explicitly considers the distance between twoneighboring SNPs, as well as the genotyping error rate and theheterozygous rate. This new method is tested on the simulatedand real data of the Affymetrix Human Mapping 500K SNP arrays.The experimental results show that the CRP method outperformsthe conventional methods based on the hidden Markov model (HMM). Availability: Software is available upon request. Contact: xzhou{at}tmhs.org Supplementary information: Supplementary data are availableat Bioinformatics online.
Associate Editor: Alex Bateman |
| |
| Keywords: | |
| 本文献已被 PubMed Oxford 等数据库收录！ |
|
